article: drew, dr, barlow, jf and cockerill, tt (2013
TRANSCRIPT
This is a repository copy of Estimating the potential yield of small wind turbines in urban areas: A case study for Greater London, UK.
White Rose Research Online URL for this paper:http://eprints.whiterose.ac.uk/82923/
Version: Accepted Version
Article:
Drew, DR, Barlow, JF and Cockerill, TT (2013) Estimating the potential yield of small wind turbines in urban areas: A case study for Greater London, UK. Journal of Wind Engineering and Industrial Aerodynamics, 115. C. 104 - 111. ISSN 0167-6105
https://doi.org/10.1016/j.jweia.2013.01.007
[email protected]://eprints.whiterose.ac.uk/
Reuse
Unless indicated otherwise, fulltext items are protected by copyright with all rights reserved. The copyright exception in section 29 of the Copyright, Designs and Patents Act 1988 allows the making of a single copy solely for the purpose of non-commercial research or private study within the limits of fair dealing. The publisher or other rights-holder may allow further reproduction and re-use of this version - refer to the White Rose Research Online record for this item. Where records identify the publisher as the copyright holder, users can verify any specific terms of use on the publisher’s website.
Takedown
If you consider content in White Rose Research Online to be in breach of UK law, please notify us by emailing [email protected] including the URL of the record and the reason for the withdrawal request.
Elsevier Editorial System(tm) for Journal of Wind Engineering & Industrial
Aerodynamics
Manuscript Draft
Manuscript Number:
Title: Estimating the potential yield of small wind turbines in urban areas: A case study for Greater
London, UK.
Article Type: Full Length Article
Keywords: wind; energy; micro-generation; urban; boundary layer; roughness length; morphology
Corresponding Author: Dr Daniel Robert Drew,
Corresponding Author's Institution: University of Reading
First Author: Daniel Robert Drew
Order of Authors: Daniel Robert Drew; Janet F Barlow; Tim T Cockerill
We derive a new map of annual mean wind speeds across Greater London
Results used to assess the best location for small wind turbine installations
Small wind turbines perform better towards outskirts of Greater London
Distance from city centre is a useful parameter for siting small wind turbines
Very few sites identified which meet threshold wind speed outlined in literature
*Highlights (for review)
1
Estimating the potential yield of small wind turbines in urban areas: A case study 1
for Greater London, UK. 2
D.R. Drewa, J.F. Barlow
a, T.T. Cockerill
b 3
a Department of Meteorology, University of Reading, UK 4
b Centre for Environmental Policy, Imperial College London, UK 5
Abstract 6
To optimise the placement of small wind turbines in urban areas a detailed understanding of the 7
spatial variability of the wind resource is required. At present, due to a lack of observations, the 8
NOABL wind speed database is frequently used to estimate the wind resource at a potential site. 9
However, recent work has shown that this tends to overestimate the wind speed in urban areas. This 10
paper suggests a method for adjusting the predictions of the NOABL in urban areas by considering 11
the impact of the underlying surface on a neighbourhood scale. In which, the nature of the surface is 12
characterised on a 1 km2 resolution using an urban morphology database. 13
The model was then used to estimate the variability of the annual mean wind speed across Greater 14
London at a height typical of current small wind turbine installations. Initial validation of the results 15
suggests that the predicted wind speeds are considerably more accurate than the NOABL values. The 16
derived wind map therefore currently provides the best opportunity to identify the neighbourhoods 17
in Greater London at which small wind turbines yield their highest energy production. 18
The results showed that the wind speed predicted across London is relatively low, exceeding 4 ms-1
19
at only 27% of the neighbourhoods in the city. Of these sites less than 10% are within 10 km of the 20
city centre, with the majority over 20 km from the city centre. Consequently, it is predicted that 21
small wind turbines tend to perform better towards the outskirts of the city, therefore for cities 22
エキIエ aキデ デエW BヴェWゲゲ IラミIWミデヴキI ヴキミェ マラSWノが ゲIエ ;ゲ GヴW;デWヴ LラミSラミが けSキゲデ;ミIW aヴラマ Iキデ IWミデヴWげ キゲ ; 23
useful parameter for siting small wind turbines. However, there are a number of neighbourhoods 24
close to the city centre at which the wind speed is relatively high and these sites can only been 25
identified with a detailed representation of the urban surface, such as that developed in this study. 26
KEYWORDS: wind, energy, micro-generation, urban, boundary layer, roughness length, morphology 27
1. Introduction 28
To reduce the carbon emissions associated with the electricity delivered to the built environment, 29
the UK government has developed a number of schemes to incentivise the growth of micro-30
generation technologies, including the Low Carbon Buildings Programme, the Code for Sustainable 31
Homes and the Feed-in tariffs Order (Allen et al., 2008; Walker, 2011). As a result there has been an 32
increase in the number of micro-generation technology installations in the UK, including micro-wind 33
turbines (Bergman and Jardine, 2009; RenewableUK, 2011). However, a number of high profile field 34
studies have shown that currently, small wind turbines installed in urban areas in the UK generally 35
produce less energy than expected prior to installation. This has raised doubts about their potential, 36
*ManuscriptClick here to view linked References
2
both in the context of the financial benefits to the owner and with respect to decarbonising the UK 37
energy supply (Encraft, 2009; James et al., 2010). 38
The literature suggests the reason for the poor performance is twofold: Firstly, the majority of the 39
turbines installed in urban areas are designed without taking into account the complex nature of the 40
wind resource at roof level. Consequently, a number of recent studies have focused on designing 41
wind turbines specifically for urban applications (Booker et al., 2010; Henriques et al., 2009; Muller 42
et al., 2009). Secondly, due to the difficulty estimating the wind resource in an urban area there has 43
been poor placement of the turbines. To optimise the placement of the turbines an accurate method 44
of assessing the variability of the wind resource across a wide urban area is required. 45
For large-scale wind turbine installations extensive wind monitoring is generally conducted to 46
identify potential sites, however, due to financial constraints, this is rarely possible for small urban 47
installations. Bahaj et al. (2007) and Allen et al. (2008) assessed the performance of small wind 48
turbines in urban areas using wind speed data collected at Met Office weather stations however 49
such data are relatively scarce in urban areas. Consequently, to identify the best sites over a wide 50
area there is a reliance on modelled wind speed data. There are several sources of wind resource 51
information available in the UK. In recent years, the DECC wind speed database and Carbon Trust 52
wind speed estimator have been the most commonly used tools. However, recent studies have 53
shown there can be large inaccuracies in their predictions, particularly in urban areas (Encraft, 2009; 54
James et al., 2010). 55
This paper aims to provide guidance for optimising the placement of small wind turbines in urban 56
areas by developing an improved method of estimating the wind resource across a wide urban area. 57
The first section discusses the tools currently used to estimate the wind resource at a potential site. 58
This is followed by a discussion of the method developed in this study. Finally, the method has been 59
applied to estimate the wind speed across Greater London, from which the best sites for small wind 60
turbines (from an energy production perspective) have been identified. 61
2. Current Methods of estimating the wind resource in urban areas 62
The DECC wind speed database has been widely used by installers and planners for a number of 63
years to identify sites for the installation of micro-wind turbines (James et al., 2010; Walker, 2011). It 64
provides estimates of the annual mean wind speed at three heights (10, 25 and 45 m) on a 1 km 65
resolution. It was produced by a mass consistent flow model, NOABL (Numerical Objective Analysis 66
of the Boundary Layer), which interpolated wind speed data from 56 weather stations across the UK 67
assuming a uniform surface (Burch and Ravenscroft, 1992). However, studies have shown that the 68
database tends to overestimate the wind speed at urban locations (James et al., 2010). At 16 of the 69
25 sites considered in the Warwick wind trials the measured wind speed was over 40% lower than 70
the NOABL prediction (Encraft, 2009). The inaccuracy of the database is indicative of the simplicity of 71
the model and in particular the lack of representation of the impact of the underlying urban surface 72
on the flow. 73
An urban surface affects the flow over a range of horizontal spatial scales: city scale (up to 10 or 20 74
km), neighbourhood scale (up to 1 or 2 km) and street scale (less than 100 to 200 m) (Britter and 75
Hanna, 2003). At the street scale, interacting wakes are introduced by individual surface obstacles, 76
hence at close proximity to buildings the nature of the flow is dependent on a number of local 77
3
surface parameters such as building size, shape and orientation. This region of the urban boundary 78
layer is known as the roughness sublayer and extends from the surface up to a height of 79
approximately 2-5 times the mean building height (Roth, 2000). 80
Blackmore (2008) used wind tunnel experiments to consider flow around a range of different 81
building designs and configurations in the roughness sublayer (i.e. on the street scale). Mertens et al. 82
(2003) and Watson and Harding (2007) performed a similar analysis using CFD simulations. The 83
results from these studies provide useful guidance as to the best location for small wind turbines 84
above a specific building or within a given street. However due to cost and time constraints, it is not 85
possible to apply this method to consider the wind speed across a wide urban area. Nevertheless, by 86
considering the flow patterns in the roughness sublayer a modified NOABL estimation tool has been 87
developed. The Micro-generation Installation Standard: MIS 3003 applies correction factors to the 88
NOABL wind speed based on turbine height and urbanisation of the site, termed NOABL-MCS (MIS, 89
2009). While this approach considers the impact of the surface on the flow in the roughness 90
sublayer, it does not consider the impact which occurs on larger scales. Consequently, James et al. 91
(2010) showed that despite the adjustment, the NOABL-MCS still generally overestimates the wind 92
resource in urban areas. 93
The region directly above the roughness sublayer is known as the inertial sublayer (ISL), which 94
extends up to a height of approximately 0.1zi, where zi is the height of the UBL. In this region the 95
flow around individual buildings is averaged out, therefore the boundary layer has adapted to the 96
integrated effect of the underlying urban surface (city scale). The wind speed in neutral conditions 97
therefore is considered to be horizontally homogeneous and increases logarithmically with height 98
戟岫権岻 噺 憲茅腔 磐権 伐 穴権待 卑 岫な岻
where U is the wind speed at a height z, 憲茅 is the friction velocity and ニ キゲ ラミ K;ヴマ;ミげゲ Iラミゲデ;ミデく 99
The roughness length, z0, provides a measure of the drag exerted on the wind by the underlying 100
surface, with a higher value indicating greater drag. When the surface obstacles are densely packed, 101
such as in an urban area, they can be considered collectively as a canopy of mean height, h. This 102
results in a vertical displacement to the wind profile, known as the displacement height, d. While 103
equation 1 is strictly only valid in the ISL, Cheng and Castro (2002) and Coceal at al. (2006) have 104
shown that it is also approximately satisfied down to the top of the canopy layer for spatially 105
averaged flow. 106
The impact of the urban surface on the flow in the ISL forms the basis on the Carbon Trust wind 107
resource assessment tool. The tool enables a user to specify their postcode and the proposed height 108
of the turbine to obtain an estimate of the annual mean wind speed. The model is based on a wind 109
climatology which has uniform validity across the country, derived from the National Climate 110
Information Centre (NCIC) dataset. This is adjusted to the hub height of the turbine assuming a 111
logarithmic wind profile and the presence of a blending height lb (Best et al., 2008). Below lb the 112
wind profile is governed by the local surface characteristics, z0local and dlocal, while above lb the wind 113
profile is governed by the effective roughness of a number of surfaces z0eff. Due to the increased 114
consideration of the impact of the surface on the flow, the Carbon Trust tool generally provides 115
more accurate predictions of the wind speed in urban areas than NOABL. However, a field trial 116
carried out by the Energy Saving Trust showed that the tool tends to underestimate the wind 117
4
resource with an error of up to 20% of the measured wind speed at some sites (Energy Saving Trust, 118
2009). 119
2.1 Internal Boundary Layer Approach 120
When flow encounters a change in surface roughness, such as a boundary between a rural and an 121
urban area, it has to adjust to the new surface characteristics. Elliott (1958) and Panofsky and Dutton 122
(1984) showed that the impact of the new surface gradually propagates upwards and a new 123
boundary layer begins to grow, called an Internal Boundary Layer (IBL). Within the IBL, the wind 124
profile is governed by the local surface characteristics, whereas above the height of the IBL, the wind 125
profile remains characteristic of the upwind surface. 126
Mertens (2003) and Heath et al. (2007) considered the growth of an IBL at the boundary between a 127
rural and urban surface to estimate the wind speed in an urban area from a reference rural wind 128
speed. However both studies assumed a uniform roughness length for the whole urban area. In 129
reality, while urban surfaces are very different from the surrounding rural surfaces, they are not 130
internally uniform. Typically, because of common use, neighbourhoods (up to 1 or 2 km) tend to 131
exhibit reasonably uniform surface characteristics (e.g. residential, industry, commercial, parkland). 132
However variability of the surface on this scale has not been considered when estimating the wind 133
speed in urban areas, therefore there is a clear need to develop a method of estimating the 134
variability of the wind resource across an urban area on a neighbourhood scale. 135
3. Development of a new wind speed estimation method 136
This study uses the IBL approach outlined in Mertens (2003) to estimate the wind speed across 137
Greater London taking into account the variability of the urban surface on a neighbourhood scale. 138
The derived wind data was then combined with the characteristics of a number of small wind 139
turbines to estimate the neighbourhoods across the city at which their energy production is 140
greatest. 141
3.1 Neighbourhood scale variability 142
To represent the nature of the surface on a neighbourhood scale, Greater London was divided into 1 143
km2 gridboxes. Each gridbox was then characterised by an estimate of z0 and d. A common approach 144
of estimating the magnitude of z0 and d over a wide area is to use land use as a proxy (Barlow et al., 145
2008; Boehme and Wallace, 2008; Rooney, 2001). However, the problem for those interested in 146
urban areas is that land use categories are usually very broad as they have to cover all types of land 147
use (Britter and Hanna, 2003). For example, there are only two urban categories (urban and 148
suburban) in the land use data used by the Carbon Trust model. A further problem with this 149
approach is the assumption that pre-determined surface parameter values are applicable to 150
different surfaces (i.e. all city centre surfaces are assigned the same z0 value). In reality, the surface 151
characteristics of one urban surface are likely to be different from that of another and consequently 152
there can be large variability in the magnitude of the surface parameters (Wieringa, 1993). 153
More precise estimates of z0 and d can be made using information about the size and spacing of the 154
buildings, this is known as a morphological approach (Britter and Hanna, 2003). This study has 155
estimated the magnitude of z0 and d using expressions derived by Macdonald et al. (1998) 156
5
権待月 噺 磐な 伐 穴月卑 崕伐 釆どの紅 系帖腔態 磐な 伐 穴月卑 膏庁挽貸待泰崗 岫に岻
穴月 噺 な 髪 畦貸碇鍋岫膏牒 伐 な岻 岫ぬ岻
where CD is the drag coefficient of a single obstacle, A is a coefficient derived from experimental 157
evidence and é is a parameter which modifies the drag coefficient to a value more appropriate to 158
the particular configuration of obstacles. For this study, these values were taken to be, é=0.55, 159
A=3.59 and CD=1.2. 160
The expressions are dependent on three building morphology parameters, h the mean building 161
height, ゜p the plan area ratio (the ratio of the total plan area of the surface obstacles to the total 162
plan area) and ゜f the frontal area ratio. These parameters were computed for Greater London on a 1 163
km2 resolution as part of the LUCID project (Evans, 2009). Even though ゜f was only derived for two 164
wind directions 180° (Southerly) and 270° (Westerly), Evans (2009) showed that the frontal area for 165
a particular wind direction is almost identical to the value for the opposite direction, irrespective of 166
building shape. Consequently, the magnitude of the roughness length calculated for westerly flow 167
z0(270) can be considered to be equivalent to that for easterly flow z0(90) (similarly z0(180) Я 168
z0(360)). 169
Figure 1 shows that the derived displacement height tends to decrease with distance from the city 170
centre. At the city centre, where the buildings are relatively tall and densely packed, the 171
displacement height peaks at a magnitude of 19.5 m, which equates to 0.8h. In the surrounding 172
suburban region, where the buildings tend to be shorter and less densely packed, the magnitude of 173
d is lower, generally between 2 and 4 m. Figure 2 shows a similar relationship is displayed for the 174
roughness length for both wind directions, z0 peaks in the city centre (1.4 m for Westerly flow and 175
1.3 for Southerly flow) and tends to decrease to a minimum value on the outskirts. Padhra (2010) 176
suggested that the symmetry of the surface parameter plots shows that the spatial structure and 177
organisation of Greater London fits the concentric ring model proposed by Burgess (1924). However, 178
the figures also show that there are some regions of low z0 and d relatively close to the city centre, 179
these areas are generally parkland (such as Hyde Park, Regents Park and Richmond Park). 180
3.2 Internal Boundary Layer Model 181
By considering the growth of an IBL at a roughness change boundary, Mertens (2003) showed that 182
the wind speed, U, in an urban area can be estimated from a reference upwind rural wind speed UA, 183
(measured at a height zA) from 184
戟岫権岻 噺 岾 釆絞 伐 穴怠権待怠 挽 峙権 伐 穴態権待態 峩峇岾 峙権凋 伐 穴怠権待怠 峩 釆絞 伐 穴態権待態 挽峇戟凋岫権凋岻 岫ね岻
where z01 and z02 are the roughness lengths and d1 and d2 are the displacement heights of the 185
upwind and downwind surfaces respectively, and 絞 is the height of the internal boundary layer given 186
by 187
6
絞岫捲岻 噺 どにぱ権待態 釆 捲権待態挽待腿 岫の岻
where 捲 is the distance from the roughness change boundary (Elliott, 1958). 188
While Mertens (2003) assumed that a single IBL grows at the rural/urban boundary, this study 189
assumes that an IBL develops at the boundary between neighbourhoods of different roughness. 190
Equation 4 is therefore used to estimate the annual mean wind speed of each gridbox for each of 191
the 4 wind directions. As the wind speed calculated using equation 4 is dependent on the distance 192
from the roughness change boundary, 捲, the wind speed was calculated at a range of 捲 values (at 50 193
m intervals from 50 に 950 m therefore n=19). The gridbox mean wind speed, U is then calculated 194
from 195
戟岫権岻 噺 な券布憲沈岫権岻津沈退怠 岫は岻
To apply the IBL formula for the first gridbox within the city boundary, a reference rural wind speed 196
is required. This has been sourced from the NOABL wind speed database; this approach is consistent 197
with that of Heath et al. (2007). For the downstream gridboxes, the mean wind speed derived for 198
the previous gridbox is used as the reference wind speed. 199
To obtain an estimate of the overall annual mean wind speed for each gridbox, the wind speed for 200
each of the 4 directions has been weighted based on the frequency of the wind from each direction 201
measured at the Met Office weather station at Heathrow between 1990 and 2011 (located on the 202
Western outskirts of Greater London at 51.479, -0.449) (UK Meteorological Office, 2012). 203
204
Figure 1 Displacement height (m) of Greater London derived from urban morphology database on a 1 km2 205
resolution. 206
7
207
Figure 2 Roughness length (m) derived from urban morphology on a 1 km resolution (a) southerly/northerly 208 flow (b) westerly/easterly flow. 209
3.3 Performance of small wind turbines 210
The model was used to estimate the annual mean wind speed at a hub height typical of that 211
recommended for a number of rooftop turbines, zhub. This was taken to be either 5 m above the 212
mean building height or 10 m for the sites at which h=0 (i.e. no buildings). To estimate the potential 213
energy production of a wind turbine at each gridbox, a Weibull distribution has been assumed to 214
represent the variability of the hourly mean wind speed. 215
The Weibull probability density function is given by 216
血岫懸岻 噺 磐倦系卑 岾懸系峇賃貸怠 結捲喧 峪伐岾懸系峇賃崋 岫ば岻
where C and k are known as the scale and shape parameters respectively. This study assumes that 217
the wind speed for each gridbox fits a Rayleigh distribution, which is a special case of the Weibull 218
distribution, which occurs when k=2 (Lun and Lam, 2000; Ramrez and Carta, 2005). 219
The annual energy production of a range of different small wind turbines was estimated at each 220
gridbox by IラマHキミキミェ デエW デヴHキミWげゲ ヮラWヴ ラデヮデ Pふぶが ふェキWミ H デエW ヮラWヴ IヴWぶ キデエ デエW WWキHノノ 221
probability density function for all of the velocities within the operating range of the turbine 222
継 噺 建豹 鶏岫懸岻血岫懸岻穴懸 岫ぱ岻塚迩祢禰貼任祢禰塚迩祢禰貼日韮
where t is the number of hours in a year. 223
Figure 3 shows the details of the 30 wind turbines considered in this study. The selection was 224
considered to represent the full range of systems currently available in the UK, both in terms of size 225
and design. The figure shows that 21 horizontal axis wind turbines (HAWTs) and 9 Vertical axis wind 226
turbines (VAWTs) with a rated power ranging from 0.056 to 9.8 kW have been considered. 227
8
228
Figure 3 Details of the swept area and the rated power of the 30 turbines used in this study. 229
4. Wind resource results 230
Figure 4 shows the predicted annual mean wind speed at zhub is generally higher on the outskirts of 231
Greater London than the city centre region. The wind speed was estimated to be highest on the 232
south west outskirts, with a value of approximately 5 ms-1
. The lowest wind speeds were predicted in 233
and around the city centre, with a magnitude of 3.3 ms-1
. However, there are some regions close to 234
the city centre with a relatively high wind speed which therefore do not fit this relationship, these 235
equate to the regions of low z0 and d values (i.e. such as Hyde Park and Richmond Park). 236
4.1 Validation 237
Ideally the mean wind speed predictions of the model would be validated with measured data. 238
However, of the 8 Met Office weather stations in Greater London with wind speed observations only 239
two are currently operational and hold wind data for a minimum of 10 years; Heathrow and 240
Northolt, both of which are located towards the outskirts of the region. For both sites, the model has 241
been used to estimate the wind speed at a height of 10 m, the results have then been compared 242
with the measured wind speed data averaged over the period 2000-2010. To have further 243
confidence in the model, the predictions have also been compared with the predictions of the 244
NOABL wind speed database and Carbon Trust tool. 245
At both sites, the method outlined in this study produces a prediction of the annual mean wind 246
speed within one standard deviation of the measured value. At the Northolt site the model 247
overestimates the measured annual mean wind speed by only 3%. In comparison, the NOABL 248
database overestimates by 27% and the Carbon Trust tool underestimates by 16%. A similar result 249
was shown at the Heathrow site, with the model overestimating the annual mean wind speed by 250
only 1% in comparison to an 18% overestimate by NOABL and 31% underestimate by the Carbon 251
Trust tool. 252
9
253
Figure 4 The annual mean wind speed, U, at zhub, based on the NOABL climatology upstream. 254
5. Implications for small wind turbines in urban areas 255
The wind map of Greater London has been used to estimate the potential annual energy production 256
of 30 small wind turbines at each of the 1 km neighbourhoods across Greater London (using the 257
method outlined in section 3.3). To allow comparison between the turbines, the performance has 258
been expressed in the form of a capacity factor. This is defined as the ratio of the actual energy 259
production in a given period, to the hypothetical maximum possible. 260
5.1 Using the new wind map to investigate the performance of small wind turbines in Greater 261
London 262
Figure 5 provides an analysis of the magnitude of the capacity factor for the 30 turbines, estimated 263
for each gridbox across the 1650 km2 of Greater London. It shows there is large variability in the 264
annual energy production of the different turbines across the city, with the HAWTs generally 265
performing better. For all 9 VAWTs the median capacity factor does not exceed 6.4%, with a mean 266
value of 4.4%. In contrast, the median capacity factor is below this value for only two of the HAWTs 267
and the mean value over the 21 turbines is 10.6%. This result is largely due to the higher cut-in wind 268
speed of the VAWTs. The figure also shows that there is not a clear trend between turbine size and 269
the predicted median capacity factor. 270
In general, the performance of the turbines across Greater London is relatively poor compared to 271
large wind turbines in open areas, with the median capacity factor exceeding 15% for only two 272
turbines (turbine 19 and 29). Further analysis showed that for all gridboxes, these two turbines were 273
predicted to produce the highest capacity factors. This suggests that, assuming the power curves are 274
accurate, of the turbines considered one of these two turbines should be selected. Figure 5 also 275
shows that there is large variability in the magnitude of the capacity factor of each turbine, 276
indicating that the turbine performance varies significantly from one location within an urban area 277
10
to another. This is also seen in figure 6 which shows the mean capacity factor averaged across all 30 278
turbines for each 1 km2. These results imply that as expected the siting of a small wind turbine is 279
important. 280
281
Figure 5 Estimated capacity factor for 30 turbines across Greater London, if installed at zhub. Median, 282 minimum and maximum values across the 1650 1 km2 neighbourhoods are represented, with the turbines 283
ordered by increasing rated power. 284
5.2 How does energy production vary with location? 285
The energy production of the turbines tends to increase with distance from the city centre, (due to 286
the increase in wind speed). To explore this relationship further the magnitude of the mean capacity 287
factor (across 30 turbines) along 15 transects (from 0 to 360° every 24°) through Greater London has 288
been considered. Figure 7 shows the mean capacity factor averaged across the 15 transects as a 289
function of distance from the city centre, as well as the minimum and maximum values. The values 290
have been normalised by the mean capacity factor for the rural gridbox at the start of each transect. 291
The figure shows that the mean capacity factor generally peaks towards the outskirts of the city 292
before decreasing to a minimum value in the city centre. A similar relationship is shown for the 293
minimum value. These results suggest that for cities which fit the Burgess concentric ring model, 294
ゲIエ ;ゲ GヴW;デWヴ LラミSラミが けSキゲデ;ミIW aヴラマ Iキデ IWミデヴWげ キゲ a useful parameter for siting small wind 295
turbines. However, figure 7 also shows that there is large variability in the maximum value of the 296
mean capacity factor across the transects; there are sites close to the city centre at which the mean 297
capacity factor is relatively high. This suggests that to identify the best sites for small wind turbines 298
(in terms of energy production) further siting parameters, such as z0 and d, are required on a 299
neighbourhood scale. 300
11
301
Figure 6 Mean capacity factor of the 34 turbines at zhub for each 1 km2 neighbourhood in Greater London. 302
303
Figure 7 Mean capacity factor of 34 turbines estimated at zhub averaged along 15 transects through Greater 304 London. The values have been normalised by the mean capacity factor at the rural site at the start of each 305
transect. 306
307
12
5.3 What proportion of sites meet the RenewableUK criteria? 308
RenewableUK guidance states that it is generally worthwhile installing a small wind turbine at a site 309
with a mean wind speed of 4-5 ms-1
(RenewableUK, 2012). This study predicts that the mean wind 310
speed at zhub exceeds the threshold of 4 ms-1
at only 28% of the 1 km2 neighbourhoods in Greater 311
London. Of these neighbourhoods, the majority are located towards the outskirts of the city, figure 8 312
shows that 50% are over 22 km from the city centre. Furthermore, the wind speed exceeds the 313
threshold at only two gridboxes within a distance of 5 km of the city centre (which correspond to 314
Hyde Park). The figure also shows that only 6% of the neighbourhoods have an annual mean wind 315
speed in excess of 4.5 ms-1
, all of which are at a distance of more than 10 km from the city centre. 316
Finally, if the threshold wind speed is taken to be 5 ms-1
, small wind turbines could only be installed 317
in two neighbourhoods in Greater London, both of which are on the western outskirts of the city. 318
319
Figure 8 Probability of finding a neighbourhoods for which the predicted annual mean wind speed at zhub 320
exceeds the threshold wind speed as a function of distance from the city centre. 321
5. Conclusions 322
Urban areas have largely been considered poor sites for small wind turbines, this conclusion has 323
generally been drawn from observations of the wind resource at point locations. However, there has 324
been little work optimising the placement of the turbines. This study has developed a method for 325
estimating the variability of the annual mean wind speed across an urban area by considering the 326
impact of the surface on a neighbourhood scale, in order to identify the best sites for small wind 327
turbine installations. 328
The method has been applied to estimate the wind resource across Greater London, UK. Due to a 329
ノ;Iニ ラa マW;ゲヴWS キミS S;デ; キミ デエW Iキデが デエWヴW エ;ゲ HWWミ ノキマキデWS ;ノキS;デキラミ ラa デエW マラSWノげゲ ヮヴWdictions. 330
However, for the two sites with wind data available, the predictions were shown to be within one 331
13
standard deviation of the measured wind speed data and considerably more accurate than both of 332
the alternative site assessment tools (NOABL and Carbon Trust tool). These results suggest that the 333
wind map developed in this study therefore presents the best opportunity to assess the 334
performance of small wind turbines in Greater London. 335
The results show that generally the wind speed across London is relatively low. Of the 1650 1 km2 336
neighbourhoods within the city, only 28% exceed the guideline threshold wind speed of 4 ms-1
at 337
turbine hub height, outlined by RenewableUK. Of these sites less than 10% are within 10 km of the 338
city centre, with the majority over 20 km from the city centre. The performance of small wind 339
turbines therefore tends to be better on the outskirts of the city, particularly in the boroughs of 340
Hounslow in the west and Bromley in the south east. Consequently, for cities which fit the Burgess 341
IラミIWミデヴキI ヴキミェ マラSWノが ゲIエ ;ゲ GヴW;デWヴ LラミSラミが けSキゲデ;ミIW aヴラマ Iキデ IWミデヴWげ キゲ ; ゲWaノ ヮ;ヴ;マWデWヴ aラヴ 342
siting small wind turbines. However, there are some regions close to the city centre at which the 343
wind speed is relatively high and these sites can only be identified by representing the urban surface 344
on a neighbourhood scale. 345
The results also show that for each neighbourhood there is large variability in the performance of 346
the different turbines, with the HAWTs generally performing better than VAWTs. Averaged across all 347
neighbourhoods in London, the median capacity factor does not exceed 6.4% for any of the VAWTs. 348
In contrast, the median capacity factor is below this value for only two of the HAWTs. 349
The approach outlined in this study, has thus far only been applied to Greater London, but could be 350
replicated for all cities for which urban morphology data is available. It could therefore provide a 351
useful tool for optimising the placement of small wind turbines in urban areas for the whole of the 352
UK. The model however does not consider the impact of individual obstacles on the flow and 353
therefore does not consider the variability of the wind speed at close proximity to buildings. The 354
results can therefore be used to identify the best neighbourhoods for small wind turbines, in terms 355
of the wind resource. However, to identify the best locations within each region, scaling factors need 356
to be applied to the wind speed to account for the impact of individual buildings on the flow. 357
Acknowledgements 358
The authors acknowledge the financial support from the Engineering and Physical Sciences Research 359
Council (EPSRC) Doctoral training grant. Thanks to Sylvia Bohnenstengel for the derived products 360
from the Virtual London dataset. 361
362
363
364
365
366
367
368
14
Figure Captions 369
Figure 1 Displacement height (m) of Greater London derived from urban morphology database on a 370
1 km2 resolution. 371
Figure 2 Roughness length (m) derived from urban morphology on a 1 km resolution (a) 372
southerly/northerly flow (b) westerly/easterly flow. 373
Figure 3 Details of the swept area and the rated power of the 30 turbines used in this study 374
Figure 4 The annual mean wind speed, U, at zhub, based on the NOABL climatology upstream 375
Figure 5 Estimated capacity factor for 30 turbines across Greater London, if installed at zhub. Median, 376
minimum and maximum values across the 1650 1 km2 neighbourhoods are represented, with the 377
turbines ordered by increasing rated power. 378
Figure 6 Mean capacity factor of the 34 turbines at zhub for each 1 km2 neighbourhood in Greater 379
London. 380
Figure 7 Mean capacity factor of 34 turbines estimated at zhub averaged along 15 transects through 381
Greater London. The values have been normalised by the mean capacity factor at the rural site at 382
the start of each transect. 383
Figure 8 Probability of finding a neighbourhoods for which the predicted annual mean wind speed at 384
zhub exceeds the threshold wind speed as a function of distance from the city centre. 385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
15
References 401
Allen, S., Hammond, G., & McManus, M. (2008). Energy analysis and environmental life cycle 402
assessment of a micro-wind turbine. Power and Energy, 669-684. 403
Bahaj, A., Myers, L., & James, P. (2007). Urban energy generation: Influence of micro-wind turbine 404
output on electricity consumption in buildings. Energy and Buildings, 154-165. 405
Barlow, J., Rooney, G., von Huenerbein, S., & Bradley, S. (2008). Relating Urban Surface-Layer 406
Structure to Upwind Terrain for the Salford Experimetn (Salfex). Boundary Layer 407
Meteorology, 173-191. 408
Bergman, N., & Jardine, C. (2009). Power from the people: Domestic microgeneration and the low 409
carbon buildings programme. Environmental Change Institute. 410
Best, M., Brown, A., Clark, P., Hollis, D., Middleton, D., Rooney, G., et al. (2008). Small-scale wind 411
energy: Part 1- A review of existing knowledge. Met Office. 412
Blackmore, P. (2008). Siting micro-wind turbines on house roofs. BRE. 413
Boehme, T., & Wallace, A. (2008). Hindcasting Hourly Wind Power Across Scotland Based on Met 414
Station Data. Wind Energy, 233-244. 415
Booker, J., Mellor, P., Wrobel, R., & Drury, D. (2010). A compact, high efficiency contra-rotating 416
generator suitable for wind turbines in the urban environment. Renewable Energy, 2027-417
2033. 418
Britter, R., & Hanna, S. (2003). Flow and dispersion in urban areas. Annual Review of Fluid 419
Mechanics, 469-496. 420
Burch, S., & Ravenscroft, F. (1992). Computer modelling of the UK wind energy resource: Overview 421
report. ETSU WN7055. 422
Celik, A., Muneer, T., & Clarke, P. (2007). An investigation into micro wind energy systems for their 423
utilization in urban areas and their life cycle assessment. Power and Energy, 1107-1117. 424
Cheng, H., & Castro, I. (2002). Near-wall flow over urban-like roughness. Boundary Layer 425
Meteorology, 229-259. 426
Coceal, O., & Belcher, S. (2004). A canopy model of the mean winds through urban areas. Quarterly 427
Journal of the Royal Meteorological Society, 1349-1372. 428
Coceal, O., Thomas, T., Castro, I., & Belcher, S. (2006). Mean flow and turbulence statistics over 429
groups of urban-like cubical obstacles. Boundary Layer Meteorology, 491-519. 430
DECC. (2008). Climate Change Act. 431
Elliott, W. (1958). The growth of the atmospheric internal boundary layer. American Geophysical 432
Union, 1048-1054. 433
16
Energy Saving Trust. (2009). Location, location, location: Domestic small-scale wind field trial report. 434
Energy Saving Trust. 435
Encraft. (2009). Warwick wind trials final report. Encraft. 436
Evans, S. (2009). 3D cities and numerical weather prediction models: An overview of the methods 437
used in the LUCID project. London: UCL Centre for Advanced Spatial Analysis. 438
Heath, M., Walshe, J. D., & Watson, S. J. (2007). Estimating the potential yield of small building 439
mounted wind turbines. Wind Energy, 271-283. 440
Henriques, J., Marques da Silva, F., Estanqueiro, A., & Gato, L. (2009). Design of a new urban wind 441
turbine airfoil using a pressure-load inverse method. Renewable Energy, 2728-2734. 442
James, P., Sissons, M., Bradford, J., Myers, L., Bahaj, A., Anwar, A., et al. (2010). Implications of the 443
UK field trial of building mounted horizontal axis micro-wind turbines. Energy Policy, 6130-444
6144. 445
Lun, I., & Lam, J. (2000). A study of weibull parameters using long-term wind observations. 446
Renewable Energy, 145-153. 447
Macdonald, R. W., Griffiths, R. F., & Hall, D. J. (1998). An improved method for the estimation of 448
surface roughness of obstacle arrays. Atmospheric Environment, 1857-1864. 449
Masson, P. (1988). The formulation of areally-averaged roughness lengths. Quarterly Journal of the 450
Royal Meteorological Society, 399-420. 451
Mertens, S. (2003). The energy yield of roof-mounted wind turbines. Journal of Wind Engineering, 452
507-518. 453
Mertens, S., van Kuik, G., & van Bussel, G. (2003). Performance of an H-Darrieus in the skewed flow 454
on a roof. Solar energy engineering, 125. 455
Microgeneration Installation Standard. (2009) MIS3003. 456
Muller, G., Jentsch, M., & Stoddart, E. (2009). Vertical axis resistance type wind turbines for use in 457
buildings. Renewable Energy, 1407-1412. 458
Oke, T. (1987). Boundary Layer Climates. London: Routledge. 459
Padhra, A. (2010). Estimating the sensitivity of urban surface drag to building morphology. University 460
of Reading: PhD Thesis. 461
Panofsky, H., & Dutton, J. (1984). Atmospheric turbulence: models and methods for engineering 462
applications. Wiley. 463
Ramrez, P., & Carta, J. (2005). Influence of the data sampling interval in the estimation of the 464
parameters of the weibull wind speed probability density distribution: a case study. Energy 465
Conversion and Management, 2419-2438. 466
17
Rankine, R., Chick, J., & Harrison, G. (2006). Energy and carbon audit of a rooftop wind turbine. 467
Power and Energy, 643-654. 468
RenewableUK. (2010). Small Wind Systems. Retrieved May 2012, from RenewableUK: 469
http://www.bwea.com/small/faq.html 470
Ricciardelli, F., & Poliimeno, S. (2006). Some characteristics of the wind flow in the lower Urban 471
boundary layer. Wind Engineering and Industrial Aerodynamics, 815-832. 472
Rooney, G. (2001). Comparison of Upwind Land-Use and Roughness Length Measured in the Urban 473
Boundary Layer. Boundary Layer Meteorology, 469-486. 474
Roth, M. (2000). Review of atmospheric turbulence over cities. Quarterly Journal of the Royal 475
Meteorological Society, 941-990. 476
Trust, E. S. (2009). Location, location, location: Domestic small-scale wind field trial report. Energy 477
Saving Trust. 478
UK Meteorological Office. (2012). MIDAS Land Surface Stations data (1853-current). NCAS British 479
Atmospheric Data Centre. 480
Walker, S. (2011). Building mounted wind turbines and their suitability for the urban scale- A review 481
of methods of estimating urban wind resource. Energy and Buildings, 1852-1862. 482
Watson, S., Infield, D., Barton, J., & Wylie, S. (2007). Modelling of the Performance of a Building-483
Mounted Ducted Wind Turbine. Journal of Physics: Conference Series. 484
485
486
Figure1Click here to download high resolution image
Figure2Click here to download high resolution image
Figure3Click here to download high resolution image
Figure4Click here to download high resolution image
Figure5Click here to download high resolution image
Figure6Click here to download high resolution image
Figure7Click here to download high resolution image
Figure8Click here to download high resolution image